Ann Navarro wrote:
> >We have come out in *support* of uniform entity name/character mappings
> >across all standards. We don't think that it's in our charter to do this
> >ourselves.
>
> That's not what I'm referring to. You may think uniform names are good, but
> I'm hearing/reading that you think it's perfectly acceptable to force
> average document authors to write internal DTD subsets to be able to write
> "Pepsi ™".
You could always use an external subset.
Of course there's probably some niggling technical issue
that would make that unworkable; the document would be
"invalid" unless the external subset also contained
a suitable collection of <!ELEMENT...> and <!ATTLIST...>
declarations, which will cause validating parsers to
reject it. Non-validating parsers of course won't
read the external subset.
[ It's always bothered me that XML parsers are either
"validating" or "non-validating". This should really
be "DTD-aware" (with validation as an extra optional
feature) and "DTD-unaware".
Maybe this can be changed in XML 1.1, not that I would
hold my breath. ]
This trick *would* work in SGML though: with
IMPLYDEF ELEMENT YES ATTRIBUTE YES, the external
subset can consist solely of <!ENTITY ...> declarations.
--Joe English
